Closed Bug 876546 Opened 12 years ago Closed 12 years ago

Enable alt+click or alt+enter to save links by default (with a hidden preference to re-disable)

Categories

(Firefox :: Keyboard Navigation, defect)

Product:
Firefox
Component:
Keyboard Navigation
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

()

Status:
RESOLVED WONTFIX

People

(Reporter: ws.bugzilla, Unassigned)

Details

Bug 713052 changed the default Alt+Click behaviour with a justification that makes very little sense. If one presses a key at the wrong time, things will *usually* go wrong. This is not a reason to disable keys. Garbage in, garbage out. It might be that the real problem is a long-standing (decade-old?) bug whereby Firefox incorrectly checks the async state of the modifier key (instead of the when-clicked state), leading to a bug whereby a click followed by Alt+action could be interpreted as an Alt click. If that's the case, then *that* is what needs fixing; it's still a problem in many other scenarios. So let's just have Alt+click back by default please.
"garbage in, garbage out", a.k.a. "blame the user", is not an argument that will get you much traction in this project. The reality is that very few users benefited from the feature, and we have evidence that it confused some users. This is quite easy to re-enable for the users who do care deeply about the feature, so we're not going to be changing course here.
Status: NEW → RESOLVED
Closed: 12 years ago
Resolution: --- → WONTFIX
Was there really sufficient evidence to justify this? If so, it's not mentioned in bug 713052. Forgive me for being skeptical, but in the discussion about removing tab groups (bug 836758) the question of "how many users are using it" came up twice, and got answered zero times, so I have my doubts about how solid the evidence used for such decisions actually is.
In the absence of hard data (which is often very difficult to get - we care a lot about user privacy and that's often fundamentally at odds with data collection on the scale necessary to get useful insight and inform decisions), we fall back to using our judgement. Given that: - keyboard shortcuts in general are not widely used (even the most common ones) - this particular keyboard shortcut is largely undiscoverable (except by accident) - there exist very few internal or external consistencies that support this particular shortcut ("Alt" isn't used as a "Save" modified anywhere else that would be relatable) - we have anecdotal evidence that this was negatively impacting people who were triggering it accidentally - we've left a mechanism for easily re-adding the feature for the minority of users who would miss it ... it's pretty self-evident that we chose the right course of action. Unfortunately, given the market share that we have (hundreds of millions of users), any time we change something we will inevitably end up upsetting some users, and so I'm sorry if this particular change upset you. Thankfully, your recourse in this situation is relatively straightforward (just flip the pref).
Sigh... This is so typical. A TINY number of people complain about a long-existing and for all they know widely-used feature and suddenly that feature is gone. Ignored are the silent majority who know nothing of the issue until suddenly POOF a feature is missing (or uselessly added) to the browser. And now when one of them stops by to express a differing opinion, they are shut down and WONTFIX'ed just because they were late to the party and "the decision has been made". ALT-clicking links to download has been in Firefox since version 1.0 and it should have been a LOT harder to remove it than it was. You say some users had problems. Okay, but you have no metrics to back this up. In fact, it looks like the entire reason this was brought up anyway was that someone was using ALT-drag to highlight text in a link (which is also a lowly keyboard shortcut). How likely is it that this feature is more commonly used than ALT-click to download? One claimed obscure feature was removed just to satisfy another equally (if not moreso) obscure feature. As for the pref to re-enable, why wasn't the option instead added to *disable* ALT-click saving instead of needing to *enable* it? If someone is having trouble with it, I understand adding an option to help that. What I don't understand is blindly (you said it yourself -- you had NO meaningful metrics) disabling a feature. And how long will the pref remain? Options to enable dead features typically don't last very long -- especially compared to prefs to *disable* current features. By the way, Chrome/Chromium also use ALT-click to download link targets. Just because you can't find a non-browser analog to a UI question doesn't mean it shouldn't be in the browser.
Reading my comment again it sounds angrier than I had intended. Frustration and disappointment, yes, but I hadn't meant to be so contentious. My apologies.
Gavin, given that any such decision, like you say, will affect a large number of people even if they are a small minority in the user base, I really wish anecdotal evidence wasn't considered sufficient for making such changes. It worries me how this decision was made. There was no mention of a user getting visibly confused in a usability study, nor of Mozilla Support or Input.mozilla.org stories. At the same time, you could bet that you've annoyed tens of thousands of people given Firefox's huge user base. Was it really so clear that it's a worthwhile change? How, exactly, does one get into a mess with Alt+Click? You have to be savvy enough to actually press Alt for some reason; given that, the extent to which you get confused is surely very limited. Sure, this was undiscoverable - so why not concentrate on making it discoverable instead? Plus, lots of things are undiscoverable until someone tells you. Undiscoverable is bad, but many _undiscoverable_ things are actually pretty good in terms of _usability_.
Gavin: > this particular keyboard shortcut is largely undiscoverable (except by > accident) Easy to fix: Add a “Download” command to the right-click menu, and put “Alt+Click” to the right of it :) The “open in new tab” should probably have a “Ctrl+Click” after it too, why doesn’t it? > we have anecdotal evidence that this was negatively impacting people who were > triggering it accidentally You have anecdotal evidence for that, but hard and incontrovertible evidence that your change has impacted people negatively who were relying on it, and the anecdotal evidence takes precedence?
(In reply to Roman from comment #6) > Gavin, given that any such decision, like you say, will affect a large > number of people even if they are a small minority in the user base, I > really wish anecdotal evidence wasn't considered sufficient for making such > changes. I didn't say "anecdotal evidence is sufficient". We rely on our judgement and experience as a development team (including UX experts) to make tradeoffs, informed by anecdotal evidence (in this particular case). > How, exactly, does one get into a mess with Alt+Click? You have to be savvy > enough to actually press Alt for some reason; given that, the extent to which you > get confused is surely very limited. Read the original bug - people use Alt+Tab to switch windows, and then click links quickly. (In reply to Timwi from comment #7) > Easy to fix: Add a “Download” command to the right-click menu, and put Discoverability of the option was not the problem we were looking to solve. > You have anecdotal evidence for that, but hard and incontrovertible evidence > that your change has impacted people negatively who were relying on it, and > the anecdotal evidence takes precedence? They're both anecdotal evidence. Some users get confused by the feature, other users like the feature. If we were designing a browser from scratch today, we wouldn't introduce that feature at all. Since that's not the situation we're in, we left it in behind a pref so that the people who really miss it have an easy way out (even if we hadn't, you'd also still be left with "install an add-on" as a recourse).
(In reply to :Gavin Sharp (use gavin@gavinsharp.com for email) from comment #8) > Read the original bug - people use Alt+Tab to switch windows, and then click > links quickly. By that argument the Ctrl-click for links (open in new tab) should be removed as well because people use Ctrl+Tab to switch tabs, and then click links quickly. (Though why they are blaming the keyboard shortcut and not themselves for poor coordination, I'm not sure). > Discoverability of the option was not the problem we were looking to solve. But you listed that as one of the reasons to eliminate the shortcut. To be honest, I'm not sure exactly what problem you _were_ trying to solve. Was it simply people people accidentally beginning a download because they tried to act more quickly than they were able? Besides, allowing the modifier keys on the keyboard to modify what clicking a link does -- that seems so self-evident as to essentially be *self-discoverable*. > They're both anecdotal evidence. Some users get confused by the feature This is my biggest problem with the argument for the removal. I don't see this claim as having been demonstrated at all. I see someone saying they don't like it when they activate the feature by accident (because they held down ALT too long), but I don't see evidence that any meaningful number of users were actually *confused* or had a poor user experience because of the feature. I see a few people with a minor annoyance. > If we were designing a browser from scratch today, we wouldn't introduce that feature at all. Why/how can you say that? I mean, Chrome is relatively new and it has the feature. Alt+click exists for the exact same reason that Ctrl+click does. Would you not include Ctrl+click either? > Since that's not the > situation we're in, we left it in behind a pref so that the people who > really miss it have an easy way out But why not leave the default and allow it to be changed by those who really felt it caused a problem? If we're considering unspecific and largely meaningless data when deciding what features to remove, consider the Google queries: 'firefox disable alt click download' has about 135,000 results -- accumulated over about last _eight years_ of the feature's lifetime. 'firefox enable alt click download' has about 1,620,000 results -- accumulated over the last _six months_ since the feature's removal. That seems to give some idea of the people using the feature versus those that find it undesirable. Of course it's all anecdotal, but what the hay. > (even if we hadn't, you'd also still be > left with "install an add-on" as a recourse). That's true, and the Firefox add-on ecosystem is a primary reason I still use the browser. Still, I don't think the ability to telling users "install an add-on" shouldn't be justification in these kinds of changes. Thank you for your willingness to discuss this issue, even if nobody has any willingness to revisit the decision.
Gavin, > Read the original bug - people use Alt+Tab to switch windows, and then click > links quickly. This is due to a bug in Firefox which causes it to think that the Alt key (or Shift or Ctrl) was pressed during the click when it actually wasn’t. You should fix this underlying problem and not merely address the symptom. Roman already mentioned that in comment 0. (Remember that windows do not actually switch until you *let go* of Alt after you press Alt-Tab.) Why not also remove tabs and require a setting or an add-on for those? I’m being serious, there /are/ users who get confused by tabs and who have no mental model of webpages remaining open “in the background”. Some cannot tell the difference between a “normal” link and a link that opens a new tab; all they notice is that sometimes the Back button stops working, and that frustrates them. By your reasoning, tabs (and the ability to have more than one website open at a time) should be an add-on, or at least off by default. Why isn’t it?
(In reply to Timwi from comment #10) > This is due to a bug in Firefox which causes it to think that the Alt key > (or Shift or Ctrl) was pressed during the click when it actually wasn’t. No, the Alt key is actually pressed in this case (the link happens quickly after the switch, before the Alt key has been raised).
(In reply to :Gavin Sharp (use gavin@gavinsharp.com for email) from comment #11) > No, the Alt key is actually pressed in this case (the link happens quickly > after the switch, before the Alt key has been raised). We may be talking different OSes here, but in Windows, that is impossible. Windows 7 started showing a preview of the new window if you pause long enough while holding Alt, but you still can't click until you let go of Alt. The quick google check performed by Nick is worrying. Perhaps the number of people affected negatively was much higher than anecdotally expected by the UX team? Perhaps it's actually worthwhile reconsidering this decision?
(In reply to Roman from comment #12) > We may be talking different OSes here, but in Windows, that is impossible. > Windows 7 started showing a preview of the new window if you pause long > enough while holding Alt, but you still can't click until you let go of Alt. That's true - I may have chosen the wrong example. Clicking a link and then switching away (i.e. pressing Alt too early) is another problematic scenario, though. Same with accidentally holding Alt in cases unrelated to window-switching. I'm not claiming that there was a strong case for disabling here. On the flip side, like I mentioned earlier, from my point of view there also isn't a strong case for having the feature to begin with (I know you disagree with this). So it's a bit of a toss-up, and given that we've now shipped the disabling and have heard relatively little in terms of opposition, we're not going to make yet another change in behavior.
> Clicking a link and then switching away (i.e. pressing Alt too early) is another > problematic scenario, though. Yes, it’s problematic because it triggers *a separate bug in Firefox* (mentioned above). Pressing Alt after releasing the mouse button *must not count as Alt-Click*, but due to the above-mentioned bug, Firefox often mistakes it for Alt-Click. Regardless, why do you think it is more important to accommodate the unbelievably far-fetched use-case in which people press Alt shortly after clicking, than to accommodate people who just want a genuine useful feature? > and have heard relatively little in terms of opposition Even I have heard lots of opposition — relatively much for such a small feature — and I’m not even involved with Mozilla. What about the 1.6 million search results returned by Nick’s Google query? You acknowledged that the three of us in this thread disagree with you on the usefulness of feature, what about these 1.6 million? How much have you heard in terms of praise for the removal of the feature by default? Has anyone actually noticed it gone, sighed in relief and said “good riddance”? Since you emphasised multiple times that we have an option to re-enable the feature, will you guarantee that this option will stay and not be removed, at least not until a reasonably bug-free and actively-maintained add-on exists to replace it? The same excuse was used against Whole Words search, but where’s our add-on for that? We can’t actually write one because the API does not expose the functionality, and nobody has been able to implement, test, debug, and persistently maintain an add-on that duplicates the entire Find functionality just for this feature. Will Alt-Click-to-Download end up in that state too?
(In reply to Timwi from comment #14) > > Clicking a link and then switching away (i.e. pressing Alt too early) is another > > problematic scenario, though. > > Yes, it’s problematic because it triggers *a separate bug in Firefox* > (mentioned above). If you're seeing this kind of bug, you should file it separately and we can investigate it there. The case I'm talking about is someone accidentally pressing Alt before they've released the click. > Even I have heard lots of opposition — relatively much for such a small > feature — and I’m not even involved with Mozilla. What about the 1.6 million > search results returned by Nick’s Google query? You acknowledged that the > three of us in this thread disagree with you on the usefulness of feature, > what about these 1.6 million? A count of Google search results for a particular query is not a useful way to gauge product feedback. The first few results refer to this issue, but the vast majority of those 1.6 million hits have nothing to do with what we're discussing (as you can easily see from skimming more than just the first page of results). > How much have you heard in terms of praise for the removal of the feature by > default? Has anyone actually noticed it gone, sighed in relief and said > “good riddance”? You can't really draw many conclusions from lack of positive feedback (particularly feature removals, where the benefits tend to be longer term and not directly linked to the actual change). People who are satisfied typically aren't as motivated to submit feedback as people who are upset, so there's an inherent bias. Whole words search certainly is possible to write as an add-on, that someone hasn't written it may be a better indication of lack of demand than of difficulty.
You need to log in before you can comment on or make changes to this bug.